JP5857682B2 - H-shaped steel flaw detection method and apparatus - Google Patents

Description

  The present invention relates to a H-shaped steel flaw detection method and apparatus for detecting surface defects present in a region to be inspected in a manufacturing process, an inspection process, and the like of H-shaped steel.

For example, a technique described in Patent Document 1 is known as a technique for detecting a surface defect generated on the surface of an H-shaped steel moving on a conveyance line in a manufacturing process.
The technique of this Patent Document 1 generates two-dimensional luminance image data by optically measuring radiation emitted from the surface of an H-shaped steel moving in a hot state, and based on the two-dimensional luminance image data. This is a technique for generating two-dimensional temperature data and detecting wrinkles generated on the surface of the shape steel based on the two-dimensional temperature data.

JP 2004-279161 A

However, when detecting the surface flaw of an H-section steel that is moving in a hot state, it is difficult to obtain highly accurate two-dimensional luminance image data due to disturbance caused by the fume generated around the H-section steel. . For this reason, it is difficult to automatically detect the surface defects of the H-shaped steel being conveyed.
Therefore, the present invention has been made paying attention to the unsolved problems of the above-described conventional example, and can detect the surface defects of the H-section steel that can be detected with high accuracy. It is an object to provide a method and apparatus.

In order to achieve the above object, the H-section steel flaw detection method according to claim 1 of the present invention is based on photographed image data obtained by photographing the flange and web of the H-section steel being conveyed in a cold state. A method for detecting defects in a H-section steel that detects surface defects existing in the inspection area of the H-section steel, wherein light is applied to the flange and the web so that halation occurs in the inspection area of the photographed image data. A region to be inspected for calculating a specific position of the web as the region to be inspected based on the captured image data and the shape data of the H-shaped steel stored in advance, and the region to be inspected A filter processing step of correcting data having a small change in shading in one of dark and light colors, and converting the corrected shading image data of the inspected area into binary data; Black data among the binary data is and a flaw recognition step determines that the surface defects in the case of having an area larger than a predetermined value.
According to this invention, since the surface defect of the H-section steel conveyed in the cold state is detected, compared with the conventional method for detecting the surface defect of the H-section steel conveyed in the hot state. Since the flange and the web are continuously irradiated with light so that disturbance such as fume does not occur and halation occurs in the captured image, the captured image data can be obtained with high accuracy.

Further, in the invention according to claim 2, the inspection area calculation step calculates the position of the vertical surface of the flange from the captured image data, reads the shape dimension of the flange from the shape data of the H-section steel, The shape dimension of the joint between the web and the flange of the photographed image data is calculated based on the position of the vertical surface of the flange and the shape dimension of the flange, and the web is identified based on the shape dimension of the joint. The position is calculated as the inspection area.
According to the present invention, the vertical surface of the flange having a relatively stable light reflectance is calculated from the photographed image data, and the inspected area of the web is set based on the calculated surface. Even if the web is transported in a state in which the web is inspected, the inspection area of the web can be reliably set.

On the other hand, the H-shaped steel wrinkle detecting device according to claim 3 has a lighting device that irradiates light to the flange and web of the H-shaped steel, and a camera that shoots a region irradiated with light by the lighting device, An apparatus for detecting defects in an H-section steel that detects surface defects existing in an inspection area of the H-section steel based on captured image data captured by the camera, wherein the illumination device inspects the captured image data. The flange and the web are irradiated with light so as to cause halation in the area , and the data with little change in shading is corrected to one of the dark color and the light color in the light and shade photographed image data of the area to be inspected. Filtering processing means for converting the grayscale image data of the inspected area into binary data, and if the black data of the binary data has an area of a predetermined value or more, it is a surface defect. It includes a flaw recognition unit constant for, a.
According to the present invention, captured image data of the surface of the H-section steel being conveyed can be obtained with high accuracy, and surface defects of the H-section steel can be automatically detected.

  According to the method and apparatus for detecting defects in H-section steel according to the present invention, surface defects of the H-section steel being conveyed in a cold state are detected, and light is applied to the flange and web so that halation occurs in the photographed image. Since irradiation is continued, surface defects of the H-section steel can be detected with high accuracy.

It is a block diagram of the device which detects the surface defect of H section steel concerning the present invention. It is the figure which showed the apparatus which detects the surface defect of H-section steel from the advancing direction of H-section steel. It is the figure which showed arrangement | positioning of the camera which comprises the apparatus which detects the surface defect of H-section steel, and an illuminating device. It is a flowchart which shows the content of the wrinkle detection process which a wrinkle detection part performs. It is a figure which shows the camera and imaging | photography image data which image | photograph the surface of H-section steel. It is a figure which shows the method of calculating the position of the vertical surface of a flange based on picked-up image data. It is a figure which shows the method of calculating the position of a junction part. It is a figure which shows the method of setting a dead zone. It is a figure which shows the method of setting a to-be-inspected area | region. It is a figure which shows the state in which the big aggregate | assembly of black data and many points | pieces (shade by a shading change) exist in the to-be-inspected area | region of picked-up image data. It is a figure which shows the state which performed the process which correct | amends the data with few shade changes in a to-be-inspected area | region from FIG. 10 image data to one of a dark color and a light color. It is a figure which shows the state converted into the binary data from the picked-up image data of FIG.

DESCRIPTION OF EMBODIMENTS Hereinafter, modes for carrying out the present invention (hereinafter referred to as embodiments) will be described in detail with reference to the drawings.
FIG. 1 is a block diagram showing a flaw detection device that detects surface defects of the H-section steel 1 being transported in a refining line that is in a cold state, and FIG. 2 shows the flaw detection device from the conveyance direction of the H-section steel 1. It is the shown cross-sectional outline.
As shown in FIG. 1, the H-section steel 1 is arranged such that the web 2 is horizontal and the pair of flanges 3 extend in the vertical direction, and is conveyed in the longitudinal direction of the arrow A direction.
As shown in FIGS. 1 and 2A, the wrinkle detection device of the present embodiment includes two pairs of lighting devices 4U and a camera 5U that are disposed above the web 2 and spaced apart in the width direction of the web 2. The image processing unit 6 includes two pairs of lighting devices 4 </ b> L and a camera 5 </ b> L that are arranged below the web 2 and spaced apart in the width direction of the web 2. The cameras 5U and 5L are area cameras.

  As shown in FIG. 3, the illumination device 4U disposed above the web 2 is disposed such that the irradiation axis 4S is inclined with respect to the vertical line by a predetermined angle θ on the downstream side in the transport direction. In addition, the camera 5U disposed above the web 2 has the photographing axis 5S in the transport direction with respect to the vertical line so that halation occurs in the photographed image photographed by the camera 5U by the light emitted from the illumination device 4U. Is inclined at the angle θ described above. And another pair of illuminating device 4U and camera 5U which are spaced apart in the width direction from this pair of illuminating device 4U and camera 5U have the same configuration.

  The illumination device 4L disposed below the web 2 is disposed such that the irradiation axis 4S is inclined by a predetermined angle θ upstream of the vertical line in the transport direction. Further, the camera 5L disposed below the web 2 has the photographing axis 5S in the transport direction with respect to the vertical line so that halation occurs in the photographed image photographed by the camera 5L by the light emitted from the illumination device 4L. Inclined by the above-mentioned angle θ on the downstream side. The other pair of illumination devices 4L and the camera 5L, which are spaced apart from the pair of illumination devices 4L and the camera 5L in the width direction, have the same configuration.

  Further, each pair of the illumination device 4U, the camera 5U, the illumination device 4L, and the camera 5L can be moved up and down by a lifting mechanism indicated by reference numerals 7 and 8 in FIG. That is, the height of the web 2 of the H-section steel 1 conveyed in the refining line and the width of the flange 3 are various, and even if the shape of the H-section steel 1 changes, each pair of lighting devices 4U and cameras 5U. In addition, the elevating mechanisms 7 and 8 raise and lower the illumination device 4L and the camera 5L, so that halation occurs in the captured images taken by the cameras 5U and 5L by the light emitted from the illumination devices 4U and 4L.

  Water droplets may adhere to the upper surface or the lower surface of the web 2 due to processing using water such as a cooling process or a cleaning process. If the photographed image is photographed without causing halation, this water drop may be erroneously detected as a wrinkle. This is because the irradiated light is irregularly reflected by the water droplets, and an apparent contrast difference is generated, thereby causing a false detection. In the present embodiment, by generating the halation, it is possible to mask only water droplets that are erroneously detected as wrinkles as described above, and to detect only the target wrinkles with higher accuracy.

In the case of the H-section steel 1 having a large web surface width, that is, a large distance between the pair of flanges 3, as shown in FIG. 2A, two cameras 5 </ b> U arranged above the web 2 are provided. Photographed image data of the upper surface of the photographed web 2 and the upper part of the pair of flanges, and imaged image data of the photographed lower surface of the web 2 and the lower part of the pair of flanges photographed by the two cameras 5L arranged below the web 2. The image is transmitted to the image processing unit 6.
Further, as shown in FIG. 2B, in the case of the H-section steel 1 having a small web surface width, the top surface of the web 2 and a pair of flanges photographed by one camera 5U disposed above the web 2 And the captured image data of the lower surface of the web 2 and the lower part of the pair of flanges captured by a single camera 5L disposed below the web 2 are transmitted to the image processing unit 6.

The image processing unit 6 includes an image capturing unit 9, a wrinkle detection unit 10, and a shape data unit 11.
When the width of the web surface of the H-section steel 1 is large, the image capturing unit 9 captures images of the upper surface of the web 2 and the upper portions of the pair of flanges captured by the two cameras 5U disposed above the web 2. Data and captured image data of the lower surface of the web 2 taken by the two cameras 5L arranged below the web 2 and the lower part of the pair of flanges are acquired at a preset image capture cycle, and the storage unit (Not shown). On the other hand, when the width of the web surface of the H-section steel 1 is small, the captured image data of the upper surface of the web 2 and the upper part of the pair of flanges captured by one camera 5U disposed above the web 2 and the web 2 The captured image data of the lower surface of the photographed web 2 and the lower part of the pair of flanges photographed by one camera 5L arranged below is acquired at a preset image capture period and stored in the storage unit. Here, the image capture cycle is, for example, a cycle in which the H-section steel 1 is conveyed 100 mm.

The shape data section 11 stores information such as the flange width and the flange thickness of the flange 3 of the H-section steel 1 for which wrinkle detection is performed.
The wrinkle detection unit 10 sequentially reads captured image data from the storage unit of the image capturing unit 9 and performs wrinkle detection processing shown in the flowchart of FIG.
In the wrinkle detection process performed by the wrinkle detection unit 10, first, inspected area calculation processing is performed in step ST <b> 1. Next, the process proceeds to step ST2 to perform filter processing. Subsequently, it transfers to step ST3 and performs an area determination process. Finally, the process goes to step ST4 to perform a result output process.

Details of the inspection area calculation process shown in step ST1 of FIG. 4 will be described with reference to FIGS.
FIG. 5A shows a pair of cameras 5U and 5U for photographing the upper surface and a pair of flanges of the web 2 of the H-section steel 1, and a pair of cameras 5L and 5L for photographing the lower surface of the web 2 and the pair of flanges. Is. As shown in FIG. 5 (a), when detecting wrinkles of the H-section steel 1 having a large web surface width, a pair of cameras 5U and 5U arranged above the web 2 and a web 2 are arranged below the web 2. A pair of cameras 5L and 5L are used, but when detecting wrinkles of the H-section steel 1 having a small web surface width, as shown in FIG. The camera 5U and one camera 5L disposed below the web 2 are used.

  FIG. 5B is photographed image data taken by one (right side) camera 5U arranged above the web 2. The background 12, the vertical surface 13 of the right flange 3, the inner surface 14 of the flange 3, The flange 15 and the joint 15 of the web 2 and the upper surface 2a of the web 2 are photographed and stored in the storage unit as photographed image data. Although not shown, the other (left side) camera 5U disposed above the web 2 and the pair of cameras 5L and 5L disposed below the web 2 also store captured image data similar to FIG. Is remembered.

In the inspection area calculation process in step ST1, first, the position of the vertical surface 13 of the right flange 3 is calculated from the captured image data (see FIG. 6). Next, the flange width and the flange thickness of the flange 3 of the H-section steel 1 in the shape data portion 11 are read, and based on these values, the position of the inner surface 14 of the flange 3 and the vertical surface 13 of the right flange 3. The joint 15 is calculated (see FIG. 7). Next, since the vicinity of the joint 15 is shielded by surface roughness and the flange 3 and the lighting devices 4U and 4L are difficult to hit, the dead zone 16 having a predetermined width H is set on the upper surface 2a side of the web 2 with respect to the joint 15. (FIG. 8). Next, a region 17 to be inspected is set at a position that does not overlap the dead zone 16 on the joint 15 side of the upper surface 2 a of the web 2.
Further, the inspection area calculation process described above is also performed on the captured image data captured by the other (left side) camera 5U disposed above the web 2 and the pair of cameras 5L and 5L disposed below the web 2. .

Next, details of the filtering process shown in step ST2 of FIG. 4 will be described with reference to FIGS.
FIG. 10 shows shaded image data taken by one (right side) camera 5U arranged above the web 2. A large collection of black data and a large number of points of black data are present in the inspected area 17. Existing. Of these, many points in the black data are shadows due to shade changes.
In the filter processing, first, density correction is performed in which data with little change in density in the density photographed image data in the inspected area 17 is corrected to one of a dark color and a light color (see FIG. 11). Next, the corrected grayscale image data in the inspection area 17 is converted into binary data (see FIG. 12).

Further, the above-described filter processing is performed on the inspection area 17 set in the captured image data captured by the other (left side) camera 5U disposed above the web 2 and the pair of cameras 5L and 5L disposed below the web 2. The same procedure is followed.
In the area determination process in step ST3, the area of the black data 18 in the region 17 to be inspected is calculated in the grayscale image data obtained from the binarization process data in FIG.

Then, in the result output process of step ST4, if the black data 18 in the inspected area 17 in FIG. 12 has an area of a predetermined value or more, it is transmitted to the host computer, or the image is output or the printer is output. To indicate “surface defects”.
In addition, the area determination process and the result output process described above, the other (left side) camera 5U disposed above the web 2, and the pair of cameras 5L and 5L disposed below the web 2 are set to the captured image data captured. The same procedure is performed for black data existing in the inspection area 17.
Here, the inspection area calculation process of step ST1 shown in FIG. 4 of the present embodiment corresponds to the inspection area calculation process of the invention, and the filtering process of step ST2 shown in FIG. 4 is the filter of the invention. Corresponding to the processing steps, the area determination processing in step ST3 and the result output processing in step ST4 shown in FIG. 4 correspond to the wrinkle recognition step of the present invention.

Next, the effect of this embodiment will be described.
Since this embodiment is detecting the surface defect of the H-section steel 1 currently conveyed in the cold state, it compares with the conventional method of detecting the surface defect of the H-section steel conveyed in the hot state. Thus, disturbance such as fume does not occur, and photographed image data can be obtained with high accuracy.
The pair of lighting devices 4U and 4U disposed above the web 2 and the pair of lighting devices 4L and 4L disposed below the web 2 include a pair of cameras 5U and 5U and a pair of cameras corresponding to each lighting device. Since the H-section steel 1 is irradiated so that halation occurs in the captured images captured by 5L and 5L, the captured image data can be obtained with higher accuracy.

Further, in the inspection area calculation process of the wrinkle detection process, first, the vertical surface 12 of the flange 3 whose light reflectance is relatively stable is calculated from the photographed image data, and based on this, the upper surface of the web 2 and Since the inspection area 17 on the lower surface is set, the inspection area 17 of the web 2 can be reliably set even if the H-section steel 1 is conveyed in a meandering state.
Further, the inspected area 17 of the web 2 performs density correction that corrects data having a small change in density among the light-captured image data in the inspected area 17 to one of a dark color and a light color. Since the grayscale image data is converted into binary data, the influence of variations in illuminance and reflectance fluctuations on the upper and lower surfaces of the web 2 is suppressed, and surface defects in the inspected region 17 are detected with high accuracy. be able to.

DESCRIPTION OF SYMBOLS 1 ... H-section steel, 2 ... Web, 2a ... Upper surface of web, 3 ... Flange, 4U, 4L ... Illuminating device, 5U, 5L ... Camera, 6 ... Image processing part, 7, 8 ... Elevating mechanism, 9 ... Image taking Insertion part, 10 ... Wrinkle detection part, 11 ... Shape data part, 12 ... Background, 13 ... Vertical surface, 14 ... Inner surface of flange, 15 ... Joining part of web and flange, 16 ... Dead zone, 17 ... Inspected area, 18 ... black data

Claims (3)

This is a H-section steel flaw detection method for detecting surface defects present in the inspected area of the H-section steel based on image data obtained by photographing the flange and web of the H-section steel being conveyed in a cold state. And
Continue to irradiate light to the flange and the web so that halation occurs in the inspection area of the captured image data ,
Based on the captured image data and the shape data of the H-shaped steel stored in advance, an inspection area calculation step for calculating a specific position of the web as the inspection area;
A filter processing step of correcting data having a small change in shading in the shaded image data of the inspected area to one of dark and light colors, and converting the corrected shaded image data of the inspected area into binary data;
And a wrinkle recognition step of determining a surface defect when the black data of the binary data has an area of a predetermined value or more. . The inspection area calculation step includes:
Calculate the position of the vertical surface of the flange from the captured image data,
Read the flange dimensions from the shape data of the H-section steel,
Based on the position of the vertical surface of the flange and the shape dimension of the flange, calculate the shape dimension of the joint between the web and the flange of the captured image data,
2. The H-shaped steel flaw detection method according to claim 1, wherein a specific position of the web is calculated as the inspection area based on a shape and dimension of the joint. It has an illuminating device that irradiates light to the flange and web of the H-shaped steel, and a camera that captures an area where the illuminating device irradiates light, and based on the captured image data captured by the camera, A H-shaped steel flaw detector for detecting surface defects present in a region to be inspected,
The illumination device irradiates the flange and the web with light so that halation occurs in the inspection area of the captured image data ,
A filter processing unit that corrects data having a small change in light and shade among the light and dark image data of the region to be inspected to one of a dark color and a light color, and converts the light and shade image data of the corrected region to be inspected into binary data;
H-shaped steel wrinkle detecting device comprising: wrinkle recognizing means for determining a surface defect when black data of the binary data has an area of a predetermined value or more .

Images